Invasive Plant Science and Management 2014 7:565–579

Successes We May Not Have Had: A Retrospective Analysis of Selected Weed Biological Control Agents in the United States

Hariet L. Hinz, Mark Schwarzla¨nder, Andre´ Gassmann, and Robert S. Bourchier*

In this paper, we describe five successful classical biological weed control agents released in the United States. For each of the five arthropod species, we compared data from prerelease studies that experimentally predicted the agent’s host range with data collected postrelease. In general, experimental host range data accurately predicted or overestimated risks to nontarget plants. We compare the five cases with insects recently denied for introduction in the United States and conclude that none of the discussed agents would likely be approved if they were petitioned today. Three agents would be rejected because they potentially could attack economic plants, and two because of potential attack on threatened or endangered plants. All five biocontrol agents have contributed significantly to the successful management of major weeds with no or minimal environmental risk. We believe that the United States may miss opportunities for sustainable and environmentally benign management of weeds using biological control if the regulatory framework only considers the risks of agents as potential plant pests and treats any host-range data regarding economic or threatened and endangered species as a binary decision (i.e., mandates rejection if there is any chance of feeding or development). As a way forward we propose the following: (1) the addition of risk and benefit analyses at the habitat level with a clear ranking of decision-making criteria as part of the U.S. Department of Agriculture Animal and Plant Health Inspection Service Technical Advisory Group’s evaluation process of biocontrol agents; (2) recognition of the primacy of realized host range data for potential agents that considers the insect’s host selection behavior instead of emphasizing fundamental host range data during release evaluations, and (3) development of formalized postrelease monitoring of target and nontarget species as part of the release permit. These recommendations may initially be advanced through reassessment of current policies but may in the longer term require the implementation of dedicated biocontrol legislation. Key words: Biological control of weeds, host range, non-target attack, risk assessment, introduction policy.

Prerelease testing of classical weed biological control plant species on which the organism can complete its life agents aims to quantify risks to nontarget species and assess cycle, whereas the realized host range comprises those plant the benefits (damage inflicted on the target weed by an species the organism actually uses under natural field agent) of introducing the biocontrol agent to a new conditions (Schaffner 2001). The fundamental host range environment. Host-range testing includes delimiting both is determined using no-choice tests in which a herbivore is the fundamental and the realized host ranges of herbivo- exposed to one plant species at a time, usually under rous arthropods. The fundamental host range comprises all confined laboratory or common garden conditions without being able to exhibit its normal host-selection behavior. In DOI: 10.1614/IPSM-D-13-00095.1 contrast, the realized host range is determined with choice * First and third authors: Head Biological Weed Control and tests under field cage or natural field conditions, in which Senior Scientist, and Assistant Director, CABI Switzerland, Rue des herbivores are able to choose from among the target weed Grillons 1, 2800 Dele´mont, Switzerland; second author: Professor of and other possible host plants. Multiple case studies Entomology, Department of Plant Soil and Entomological Sciences, demonstrate that a herbivore’s realized host range under University of Idaho, Moscow ID, 83844-2339; fourth author: field conditions is always narrower than its fundamental Research Scientist, Agriculture and Agri-Food Canada, Lethbridge, host range (e.g., Blossey et al. 2001; Center et al. 2007; Canada. Corresponding author’s E-mail: [email protected] Cristofaro et al. 2013; Dudley and Kazmer 2005;

Hinz et al.: Successes we may not have had N 565

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 McFadyen et al. 2002; Medal and Cuda 2010; Paynter et representing federal agencies in the United States, other al. 2004; Pratt et al. 2009; see additional examples in organizations, and the countries of Mexico and Canada, Wapshere 1989). To our knowledge, there is not a single that advises USDA APHIS, but the insect was not case using appropriate host-range testing protocols where approved for release. The reason given for the negative the opposite result has been demonstrated. decision by USDA APHIS was that ‘‘… this insect may Recently, decisions about releases made by the U.S. negatively impact a United States crop plant (safflower)’’ Department of Agriculture Animal Plant Health Inspection and was likely based on data showing that under no-choice Service (USDA APHIS, the regulatory agency responsible conditions, the weevil’s fundamental host range included for biological control introductions) have been weighted safflower (Carthamus tinctorius L.) (Smith 2007), a crop more on the fundamental host range of a potential agent plant grown for oil production. However, in tests and seemingly ignore the realized host range. This examining the realized host range of the weevil, safflower conservative approach is a safe one, because there are no was not attacked during five open-field experiments (three known examples of agents that attack plants outside their included in the petition and two conducted subsequently fundamental host range (Paynter et al. 2004, Pemberton (Cristofaro et al. 2013; Rector et al. 2010; Smith et al. 2000, van Klinken and Edwards 2002). However, here we 2006). In addition, C. basicorne has never been reported argue that implementing it as the sole criterion for release from safflower in their shared native Eurasian range decisions significantly overestimates potential risks associ- (Cristofaro et al. 2013). It is unclear whether or how ated with the release of candidate agents. Doing so limits realized host-range data for the weevil were considered in the number of agents available for release, and consequent- the decision process or whether these data were weighed ly reduces the prospects for successful control (van Wilgen against the fundamental host range data (L. Smith, et al. 2013). personal communication). The decision to deny the release To present our arguments, it is first important to of C. basicorne was unexpected, since the guidelines in the understand the biology of host use under no-choice and TAG reviewer manual explicitly emphasize consideration choice conditions. A plant species might be able to support of realized host range data (USDA 2000, p 4-3-5). In development of a potential agent under no-choice contrast, the decision appears more reflective of a zero- conditions but not be attacked in the field for several tolerance standard in respect to any potential utilization reasons. These include but are not limited to the following: by a candidate agent on agriculturally or economically life-history traits or the host selection behavior of the agent used plant species under no-choice laboratory conditions. that typically act as filters (e.g., visual or olfactory host This case exemplifies the uncertainty that exists among location and identification cues that are bypassed under biocontrol scientists regarding the definition and quanti- confined conditions), the availability of preferred host fication of risks associated with biocontrol introductions plants (i.e., the target weed or weeds), the ability of the that are applied during the regulatory decision process. To agent to disperse and forage for preferred host plants, and a our knowledge, there are currently no guidelines for risk low probability for the agent to encounter nontarget plant assessments, though these would be important if the species. Low encounter rates may result from asynchronies petitioner is to define additional research in order to of phenologies of the nontarget plant and the biological address concerns raised by regulatory authorities during the control agent or little or no overlap between geographic introduction process. distributions or habitat requirements of nontarget plant Here, we argue that focusing solely on the fundamental species and the biological control agent, or a combination host range of potential agents, though safe, will reduce the of both (Briese 2004; Schaffner 2001; van Klinken and benefits of biological control as a management option for Edwards 2002; Wapshere 1989). The fundamental host invasive plant control. We first present an overview of the range of an agent is a genotypic trait that can relatively regulatory process in the United States. Then, to support easily be determined experimentally, whereas the realized our argument, we present five case studies of successful host range is how the fundamental host range is expressed arthropod biological control agents of herbaceous weeds within a particular environment and is therefore influenced in rangeland, wetlands, and cropping systems in North by specific environmental conditions, such as the relative America that likely would not be permitted for release availability of hosts (Louda et al. 2005b; Sheppard et al. today based solely on their fundamental host range. Our 2005; van Klinken and Edwards 2002). selection criteria were that (1) the released agent is, at least A recent case highlights the importance of considering locally, successfully controlling the target weed and (2) after the biological differences between fundamental and realized a review of host-range testing and field data, we believe that host ranges. The weevil Ceratapion basicorne (Illiger) was the agent would not be permitted for release if petitioned petitioned for release to control Centaurea solstitialis L., today although it can be considered environmentally safe. yellow starthistle. It was recommended for release by the This paper does not attempt to review all existing cases that Technical Advisory Group (TAG), an expert committee may fit the above criteria, but instead uses the selected case

566 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 studies to discuss criteria applied during the permitting protection of threatened and endangered species from any process of classical weed biocontrol agents in the United potential adverse effects and thus it has the authority to States. For each of the selected cases, we address the oppose the introduction of a weed biological control agent question of whether the agent caused any predicted or if there are reasons for concern. unpredicted nontarget attack on other plant species by Researchers develop and submit petitions for permits to comparing results of prerelease studies (predicted host release biological control agents of exotic weeds. USDA range) with postrelease studies (realized host range) using APHIS provides guidance on petition format to ensure that all available data. We then try to provide rationales all required information is included and meets minimum outlining why we believe that the agents would not be quality standards (USDA 2000). The petition summarizes released if petitioned today. The cases are described from target weed and agent biology, agent host range, and the oldest to most recent release date of respective agents. We expected impact of the agent on the target weed and summarize commonalities of the case studies, discuss identifies any potential adverse effects to the environment changes in safety standards and risk perception and provide that could arise as a result of the proposed release. The recommendations that would assure environmental safety petition is submitted to the TAG for Biological Control standards of the regulatory process for biological control Agents of Weeds, a committee convened by USDA APHIS agent introductions in the United States while at the same with representatives appointed to represent the interests of time ensuring classical biological weed control as a viable 17 cooperating agencies, countries, and organizations, management strategy for invasive plants. including most U.S. federal agencies involved in land management (Cofrancesco and Shearer 2004). The mem- bers of the TAG, who may also be biocontrol practitioners, Overview of the Current Regulatory Process for the review petitions themselves and can also exercise the option Introduction of Weed Biological Control Agents in of seeking further input from intra-agency subject experts. the United States Reviews are then discussed and summarized by the chair of The federal agency responsible for coordinating the the TAG and a recommendation for or against the release of assessment of petitions for permits to import and release the agent is forwarded to USDA APHIS. USDA APHIS nonnative herbivores as biological control agents of noxious considers the recommendation from the TAG during its weeds, and subsequently issuing those permits, is USDA decision-making process, although the agency has no APHIS, operating under the regulatory authority contained obligation to do so. If the TAG recommends the release of in the Endangered Species Act of 1973 (16 U.S.C. 1531 et. the agent, the next stage is mandatory consultations with seq.), National Environmental Policy Act of 1970 (42 tribal nations and the USFWS (outlined under Section 7 of U.S.C. 4321 et seq.), Plant Protection Act of 2000 (7 the Endangered Species Act). If USDA APHIS has cause to U.S.C 7701 et. seq.), Coastal Zone Management Act believe that issuing a permit that will allow for the of 1972 (16 U.S.C. 1451 et. seq.), and Executive Order importation and subsequent environmental release of an 13112 (64 CFR 6184) for Invasive Species of 1999. In the agent poses a risk to U.S. agricultural interests (according to United States, all movement, including the importation the Plant Protection Act), or to federally listed threatened or and release into the environment of biological control endangered species (according to the Endangered Species organisms of noxious weeds, is regulated by the Plant Act), then the agency is legally obliged to decline approval of Protection Act, which also provides the Secretary of the permit. If the petition fails to garner support during Agriculture with the authority to regulate ‘‘any enemy, mandatory tribal or USFWS consultations, the review antagonist, or competitor used to control a plant pest or process will typically be aborted. For petitions receiving noxious weed.’’ support of the TAG, USDA APHIS, tribal nations, and the Procedures outlined in the National Environmental USFWS, USDA APHIS will provide an Environmental Policy Act need to be followed primarily because issuing a Assessment (EA) as is required by the National Environ- permit allowing for the introduction and release of a mental Policy Act. If the agency concludes that nationally biological control agent constitutes an action by a federal there is a finding of no significant impact to the agency (USDA APHIS) that has the potential to negatively environment, a permit applying to all 48 mainland states impact the human environment. In addition, the Endan- will be issued to legally allow for the release of the agent. gered Species Act applies to weed biological control agent introductions because it is possible that the introduction of Case Studies a nonindigenous plant herbivore could potentially pose a risk or have direct or indirect adverse effects on federally Hypericum perforatum (L.), St. Johnswort. St. Johnswort listed threatened or endangered species or habitats (16 is a perennial herb of Eurasian origin introduced to U.S.C. 1531 et. seq.). The U.S. Department of Interior’s Australasia and North America, where it developed into a Fish and Wildlife Service (USFWS) is charged with the major weed in prairie ecosystems, rangelands, cropping

Hinz et al.: Successes we may not have had N 567

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 systems, and open forests (Jensen et al. 2002; Piper 2004). sympatrically with H. perforatum, has remained common Between 1945 and 1979, seven insect biological control in those Californian communities to which it is native agents were released for the control of St. Johnswort in (Andres 1985). Although H. calycinum was in some North America (Julien and Griffiths 1998). The leaf beetle instances severely defoliated, reports of nontarget attack Chrysolina quadrigemina Suffrian was introduced in 1945. declined with time (Andres 1985), probably because of a Both adults and larvae feed externally on the leaves and can decrease in the target weed density and subsequent beetle defoliate plants (Piper 2004). The beetle proved to be the abundance. Thus, despite the nontarget attack, which was most important single agent that contributed to the predictable from experimental testing, the introduction of successful control of St. Johnswort. In California alone, C. quadrigemina contributed greatly to the control of the infestations of more than 800,000 ha (2 million acres) were target weed and had no measurable long-term negative reduced to less than 1% of that area (Huffaker 1967; effect on either the native Hypericum species or the Huffaker and Kennett 1959), accruing more than $20 horticultural industry. million in savings for the agricultural industry between Considering only the original host-specificity data 1953 and 1959 (DeBach and Rosen 1991). In most arid presented C. quadrigemina would not be permitted for regions in south-central British Columbia, C. quadrigemina release today, simply because none of the native congeners reduced the weed to less than 2% of the prerelease levels, of H. perforatum were tested. When considering the test and it was effective in reducing the weed to negligible levels data collected after the release of the beetle in the United at release sites in southern Ontario (Jensen et al. 2002). States, it is doubtful that a release permit for C. Prerelease host-specificity tests for C. quadrigemina quadrigemina would be granted, because larvae developed concentrated mostly on screening economic plant species equally well on at least one commercially grown Hypericum in distantly related families (Currie and Garthside 1932; species and several native Hypericum species as on the target Holloway 1948). However, before introduction into weed. In addition, all nontarget Hypericum species tested Canada in 1952, additional tests were conducted with were accepted to some degree for oviposition under open- seven species in the genus Hypericum: five natives and two field conditions. A recent report reviewing the biological introduced ornamentals (Smith 1958). Feeding occurred control program against St. Johnswort in New Zealand on all Hypericum species, and C. quadrigemina was able to came to the same conclusion (Groenteman et al. 2011), complete its life cycle on two of the natives (Hypericum i.e., that C. quadrigemina (and its congener Chrysolina frondosum Michx. and Hypericum moserianum Luquet ex hyperici Fo¨rster) would not have been approved for Andre´) and on the introduced ornamental Hypericum introduction into New Zealand under current risk calycinum L. used extensively in landscaping (Smith 1958). assessment protocols in that country. The authors write: It therefore came as no surprise when, in the 1970s, the ‘‘Had this happened, NZ would have missed out on one of beetle was found attacking H. calycinum and the native its greatest biocontrol success stories, despite there being no North American Hypericum concinnum Benth. in several evidence for subsequent impact on the populations of counties in California (Andres 1985). indigenous congeners.’’ In subsequent postrelease studies with H. calycinum, H. concinnum, and two additional native North American Convolvulus arvensis L., Field Bindweed. Field bindweed species, Hypericum anagalloides Cham. & Schlecht and is a perennial vine native to Eurasia that was introduced Hypericum scouleri Hook. subsp. scouleri [5 Hypericum into North America during the 18th century. The plant can formosum Kunth var. scouleri (Hook.)], beetles were able to form dense, tangled mats that may reduce crop production complete development on all four nontargets in no-choice by as much as 60% (Littlefield 2004). In 1998, U.S. crop laboratory tests; H. concinnum and H. scouleri supported losses due to field bindweed infestations were estimated to similar adult emergence rates as H. perforatum (Andres exceed $377.8 million annually (Boldt et al. 1998). In 1985). An impact experiment also showed that larval 1989, the gall-forming mite Aceria malherbae Nuzzaci was feeding by C. quadrigemina can reduce plant size of H. introduced to North America for the biological control of concinnum. And finally, two of the native Hypericum field bindweed and 4 yr later, in 1993, also for control of species (H. concinnum and H. anagalloides) and to a lesser hedge bindweed, Calystegia sepium (L.) R. Br. Nymphs and extent, H. calycinum, were accepted for oviposition in an adults of the mite suck on plant sap, leading to gall open-field test (Andres 1985). formation on leaves and stem tips, which stunts plants and However, C. quadrigemina has not been reported reduces flowering (Littlefield 2004). attacking H. anagalloides and H. scouleri in the field, Quantitative field data on the effectiveness of A. which might be due to different habitat preferences of these malherbae are limited. In Texas, bindweed biomass has two natives and the target weed H. perforatum (Andres been reduced by over 95% in many areas, primarily on 1985). Despite reports of nontarget attack by C. roadsides, highway medians, and yards; there has been less quadrigemina, H. concinnum, which commonly occurs success in cultivated row crops (Smith et al. 2010). At some

568 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 release sites in Wyoming, bindweed populations have Since nearly all of the native Calystegia species on which decreased to innocuous levels within 4 to 5 yr following the gall mite is able to develop are geographically limited to the release of the mite (J. L. Baker, personal communi- the state of California, which has denied any release of the cation). In Oregon, the mite is spreading and reducing mite, no nontarget effects on Calystegia species have been bindweed biomass by 90% (Smith et al. 2010). In reported there to date. This result demonstrates the Canada, the mite has also shown good biological control potential value and effectiveness of geographical restrictions potential; however, there is considerable variation among on releases for some agents where dispersal is limited. release sites with regard to levels of establishment and impact, which may partly be explained by differences in Lythrum salicaria L., Purple Loosestrife. Purple loose- environmental conditions (McClay and De Clerck-Floate strife is a perennial wetland plant introduced from Eurasia 2002a). Completely galled plants are severely stunted to North America in the early 1800s. It is capable of where the mite is well established (McClay and De forming continuous stands that can displace native Clerck-Floate 2013). vegetation with significant ecological impacts on wetlands In prerelease studies conducted under quarantine (see references in Blossey et al. 2001). Five biological conditions in the United States, a total of 43 test plant control agents were released against purple loosestrife species were evaluated, 22 in the family Convolvulaceae between 1992 and 1994. Among these, two defoliating leaf and the remaining species in 20 different plant families beetles, Galerucella calmariensis L. and Galerucella pusilla (Rosenthal and Platts 1990). The test plants included Duftschmid, are particularly effective. Both larvae and two threatened and endangered listed species: Calystegia adults feed on leaves and buds and can defoliate plants peirsonii (Abrams) Brummitt, and Calystegia stebbinsii (Piper et al. 2004). The biomass at several purple loosestrife Brummitt, and four subspecies of other Calystegia spp., stands in Oregon and Washington has been reduced by which were considered rare or endangered. The mite was 90%; after several years of repeated defoliation, plant size able to reproduce and sustain populations on all nine decreased while plant mortality increased (Piper et al. Calystegia species tested, including the above-mentioned 2004). In Manitoba, nearly 100% control of purple federally listed species (Rosenthal and Platts 1990). loosestrife has been achieved at many release sites (Lindgren Considering these results, it is highly unlikely that the et al. 2002). In Michigan, G. calmariensis caused 100% gall mite would be permitted for release in 2014 based on defoliation of purple loosestrife, which resulted in reduced the mandated responsibility of the USFWS to protect stem heights by up to 85% and reduced plant cover by up threatened and endangered listed species under the to 95% accompanied by an increase in nontarget plant Endangered Species Act. Rosenthal and Platts (1990) species richness (Landis et al. 2003). argued that ‘‘Benefits from the possible control of C. Prerelease studies were conducted with 47 test plant arvensis in North America must be contrasted with any species in Europe (Blossey et al. 1994) and with 15 species harm due to possible attacks on native plants’’ and in containment facilities in the United States (12 of which regarding the potential attack on threatened and endan- were identical with species tested in Europe) (Kok et al. gered species, that ‘‘such species that occur at low levels of 1992). Under no-choice conditions, both Galerucella abundance are less subject to attack because of the density species fed and laid eggs on the native North American dependent manner in which biological control agents Lythrum alatum Pursh. (winged loosestrife) and Lythrum work. Thus, there is relatively little chance of actual californicum Torr. & Gray, the European Lythrum encounter and impact on rare Calystegia species and hyssopifolia L., the ornamental Lagerstroemia indica L. subspecies.’’ (crepe myrtle), and the native North American Decodon In subsequent experiments with one native Convolvulus verticillatus (L.) Ell. (swamp loosestrife). In no-choice larval and three native Calystegia species in field plots between transfer tests, both beetle species developed to adulthood 2005 and 2007, galling was observed on all plants, on L. alatum and the European Lythrum virgatum L. although the severity and frequency of galling was much (which is taxonomically very similar to L. salicaria). In lower on the Calystegia species when compared to C. addition, G. calmariensis successfully developed on the arvensis, and live mites were only found on two of the three native North American Ammannia auriculata Willd. and native species exposed (R. Hansen, personal communica- Ammannia latifolia L. during tests conducted in Europe. tion). These tests were conducted by attaching mite- Two open-field tests were conducted with L. alatum and infested field bindweed to test species, constituting a D. verticillatus in Europe prior to release (Blossey et al. no-choice test under field conditions. The results also 1994). In the first test, plants were exposed to beetles that suggested that the mites may not be able to successfully had overwintered and were beginning to lay eggs. Under overwinter on the nontarget plant species in Colorado or these conditions, no feeding or oviposition was recorded on that aerially dispersing mites in spring did not successfully the two test plant species. In the second test, plants were colonize nontarget plants (Smith et al. 2010). exposed to newly emerged adults of the F1 generation.

Hinz et al.: Successes we may not have had N 569

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 Under these conditions, all purple loosestrife control plants releases had already taken place in other states, which were completely defoliated and an estimated 30 to 40% of enabled field experiments assessing the realized host range. the leaf surface of the two nontarget plant species was Today, the damage to a widely used ornamental and consumed (Blossey et al. 1994). concerns of the State of California about release would Based on these results, it was predicted that limited probably prevent a release permit for the Galerucella spp. in nontarget attack by newly emerging F1 adults might occur, the United States. but that attack would be restricted to L. alatum and D. verticillatus (Blossey el al. 1994). Subsequent postrelease Linaria spp., Toadflaxes. Yellow toadflax, Linaria vulgaris monitoring determined that the mass emergence of newly Miller, and Dalmatian toadflax, Linaria dalmatica (L.) emerged F1 Galerucella adults also resulted in unpredicted, Miller (syn: Linaria genistifolia subsp. dalmatica), were localized, short-term (temporal) attack on at least two introduced as ornamentals from eastern Europe in the completely unrelated plants (Rosa multiflora Thumb. and second half of the 19th century and they, along with Potentilla anserina L.), but this was transitory in nature and hybrids (Ward et al. 2009), are now widespread across disappeared as loosestrife populations declined (Blossey et North America (Nowierski 2004). Linaria species are al. 2001). As predicted, similar transitory adult feeding also avoided by cattle and reduce the productivity of infested occurred on the two closely related native North American rangelands. Yellow toadflax is also a problem in cultivated species, L. alatum and D. verticillatus (Blossey et al. 2001; lands that has increased with the adoption of low-till Corrigan et al. 1998). A few egg masses were found on L. cultivation practices (McClay and De Clerck-Floate alatum, but no late instars were observed to develop on this 2002b). Between 1962 and 1996, six biological control plant in the field (Corrigan et al. 1998), in contrast to agents were released for the control of Dalmatian toadflax, results from larval-transfer tests in the laboratory reported yellow toadflax, or both in North America (De Clerck- above. Floate and Harris 2002; Wilson et al. 2005). A recent The California Department of Food and Agriculture molecular study (Tosˇevski et al. 2011) has shown that the (CDFA) initially refused to permit the release of beetles in stem-boring weevil Mecinus janthinus Germar, first released California because no-choice test results in quarantine (Kok in North America in 1991, is actually composed of two et al. 1992) showed substantial feeding and oviposition closely related species, M. janthinus from L. vulgaris and (especially by G. calmariensis)onLagerstroemia indica the cryptic Mecinus janthiniformis Tosˇevski & Caldara (crepe myrtle), an economically important nursery plant from the L. genistifolia–L. dalmatica complex in southeast- and popular ornamental. The CDFA asked biocontrol ern Europe. Mecinus janthiniformis is now credited with researchers in Oregon to clarify the potential for nontarget successfully controlling Dalmatian toadflax in the north- damage to this plant. Schooler et al. (2003) established a western United States and British Columbia (Schat et al. replicated field experiment in 1996 (4 yr after release of the 2011; Van Hezewijk et al. 2010; Weed and Schwarzla¨nder beetles), in which L. indica and purple loosestrife were 2014). The larvae of this weevil mine in the stems and in placed at increasing distances from a purple loosestrife spring, emerging adults feed on host-plant shoot tips, stand heavily attacked by Galerucella spp. Extensive leading to a complete suppression of flowering and seed defoliation of L. indica was limited to within 30 m (98 ft) production, and severe stunting of shoots (De Clerck- and approached zero at about 50 m from the beetle front. Floate and Harris 2002). Recent studies on the landscape At each distance, damage was lower for L. indica than for scale showed that toadflax patches decreased in density and purple loosestrife. The authors concluded that the two became more fragmented over time, with 15% of patches Galerucella species would only have minimal, short-term disappearing completely due to M. janthiniformis (Van effects on populations of crepe myrtle because the plant is Hezewijk et al. 2010) and that the weevil had a direct unsuitable for beetle development, has little geographic negative effect on the stem-density growth rate of L. overlap with purple loosestrife, and non-target attack dalmatica patches (Weed and Schwarzla¨nder 2014). decreased with increasing distance from purple loosestrife Recently, large population increases of M. janthinus and stands harboring beetles. Based on these results, both reductions in toadflax densities have been reported at some Galerucella species were permitted for release in northern sites of L. vulgaris in Canada and the United States (A. California. Until 2010, the beetles established at least at McClay, J. Milan, and S. Sing, personal communications). one location in northern California where no crepe myrtle Prerelease studies with M. janthiniformis (then named occurred (M. Pitcairn, personal communication). Once M. janthinus) collected from L. genistifolia in Macedonia beetles disperse to more southern locations where crepe were conducted in Europe with 38 test plant species, 22 in myrtle does occur, follow-up surveys will be conducted (M. the former family Scrophulariaceae (Jeanneret and Schroe- Pitcairn, personal communication). der 1992), which has recently been revised (Albach et al. The experiments leading to the final approval for release 2005). A recent comprehensive study on the phylogeny of the beetles in California were only possible because of toadflaxes suggests that the native North American

570 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 Nuttallanthus is most probably nested within the genus moved out of the release location, and no attack has been Linaria (Fernandez-Mazueco et al. 2013). The emphasis of seen on any Sairocarpus (5 Antirrhinum) spp. at that the original test-plant list was on ornamentals, and only location (L. Smith, personal communication). However, nine native North American species were tested, one in the M. janthiniformis has recently been sighted in northern tribe Antirrhineae (to which the target weed belongs) and California, most likely immigrating inadvertently from eight in other tribes. Jeanneret and Schroeder (1992) populations in southern Oregon (Villegas et al. 2011). concluded that the host range of M. janthiniformis (then In 2014, M. janthiniformis would not be approved for named M. janthinus) was restricted to the genus Linaria release. With advances in host-range testing methodologies and did not develop on snapdragon, Antirrhinum majus L., and inclusion of native species in test lists, the test list used in an important ornamental in North America. These results, the 1980s would be considered incomplete, and justifiably however, were based primarily on adult feeding and so. Subsequent testing on additional species, mostly native to oviposition tests using cut shoots and on a limited number North America, indicated that the fundamental host ranges of larval transfer tests (Jeanneret and Schroeder 1992). of M. janthinus and M. janthiniformis comprise several Subsequent no-choice tests with additional native plant nontarget plants, all members of the same tribe, Anti- species in 2000 and from 2010 to 2011 using potted rrhineae, as the targets L. vulgaris and L. dalmatica.Given plants, have shown that M. janthinus and M. janthiniformis that the fundamental host range of M. janthiniformis also are able to develop to the adult stage on two native species includes a state-listed rare species (N. canadensis in Ohio), a in tribe Antirrhineae, i.e., Sairocarpus virga (A. Gray) D.A. release permit would likely be denied today. Additional Sutton (5 Antirrhinum virga A. Gray) and Nuttallanthus testing under multiple-choice conditions and postrelease canadensis (L.) D.A. Sutton [5 Linaria canadensis (L.) monitoring is required to determine the realized host ranges Mill.], which is listed as endangered in Ohio (Tosˇevski et of both Mecinus species. al. 2012). Mecinus janthinus collected from L. vulgaris developed to the mature larval stage on three additional Solanum viarum Dunal, Tropical Soda Apple. Tropical native species: Sairocarpus nuttallianus (Benth. ex A. DC.) soda apple is native to South America and was first D.A. Sutton, Neogaerrhinum strictum (Hook. & Arn.) reported in southern Florida in 1988 (Mullahey et al. Rothm. (5 Antirrhinum kelloggii Greene), and Mauran- 1993). Since then it has spread very rapidly throughout della antirrhiniflora (Humb. & Bonpl. ex Willd.) Rothm. Florida and into other southern states, where it invades (5 Maurandya antirrhiniflora Humb. & Bonpl. ex Willd.) rangelands, pastures, and natural areas (Mullahey et al. (Gassmann et al. 2002; Tosˇevski et al. 2012). A population 1996). Dense patches of S. viarum reduce cattle stocking of Mecinus from Russia, which could have been M. rates and limit their movement, and serve as reservoirs for janthinus or M. janthiniformis, also developed on Sair- pests of Solanaceae crops (Diaz et al. 2014). Exploration ocarpus multiflorus (Pennell) D.A. Sutton (5 Antirrhinum for biological control agents of S. viarum began in 1994 in multiflorum Pennell) under no-choice conditions (Gass- its area of origin and in 2003, the leaf-feeding beetle mann et al. 2002). In a postrelease multiple-choice field Gratiana boliviana Spaeth was introduced into Florida cage test conducted in Bozeman, MT, in 2001 with insects (Medal and Cuda 2010). Both larvae and adults feed on that were most probably M. janthiniformis, where S. virga, the leaves of S. viarum and can completely defoliate plants, S. multiflorus, and N. strictum were exposed, only S. virga resulting in reduced growth and fruit production (Diaz et supported the development of a single adult Mecinus. The al. 2014; Overholt et al. 2009). Up to a 90% decline in the mean offspring per stem was 0.06 on S. virga vs. 0.95 on L. density of S. viarum has been observed in high-density dalmatica (Gassmann et al. 2002), suggesting that in populations only 2 to 3 yr after beetle releases (Overholt et presence of the target weed, S. virga is not a preferred host. al. 2010). Percentage of cover of S. viarum decreased from In a postrelease evaluation of the potential for nontarget an initial level of 80 to 90% to 5 to 10%, 2 yr after release attack by M. janthiniformis, only feeding and oviposition (Medal et al. 2008). A preliminary economic analysis scars, but no development, were found on the native suggests that the release of G. boliviana reduced the Nuttallanthus texanus (Scheele) under no-choice conditions management costs of S. viarum by approximately 50% (Breiter and Seastedt 2007). statewide, resulting in potential statewide savings of $3 to 8 Initially, the State of California refused to permit release million annually (Salaudeen et al. 2012). of M. janthiniformis (named at that time M. janthinus), During prerelease multiple-choice host-specificity tests, because of concerns about native Scrophulariaceae in that 123 plant species (including 25 species in the genus state. However, in 2008, a release of 1,400 M. janthini- Solanum) in 35 families were exposed to G. boliviana under formis weevils was made in California in a rapidly quarantine conditions (Medal et al. 2002). Adults fed on expanding infestation of L. dalmatica nearly 644 km four test species and laid eggs on two: the introduced weed (400 mi) distant from the closest known location of S. virga Solanum torvum Swartz and the crop Solanum melongena L. (Villegas 2009). Results to date are that the weevil has not (eggplant). An average of 68 eggs were found on S. viarum,

Hinz et al.: Successes we may not have had N 571

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 but only 0.2 eggs were found on eggplant, suggesting it is a 1998; Schooler et al. 2003), or both. The only case in which marginal host (Medal et al 2002). In no-choice larval- nontarget attack occurred on plant species unrelated to the transfer tests, no development occurred on any of the 21 target weed that was not predicted prerelease occurred after nontarget test plant species exposed under quarantine outbreaks of the biological control agent, and these effects conditions in Florida, but a few larvae did develop to the were only transitory (i.e., Galerucella spp. for purple adult stage on eggplant in tests conducted in Argentina loosestrife; Blossey et al. 2001). For cases in which nontarget (Medal et al. 2002). This triggered a field survey of 30 attack persisted, there are no reported population-level unsprayed eggplant fields in the native range of G. effects (C. quadrigemina on H. concinnum; Andres 1985), boliviana (Argentina, Brazil, Paraguay, and Uruguay) and the ornamental H. calycinum is still grown and between 1997 and 2000 (Medal et al. 2002). No larvae marketed. The other two agents discussed (A. malherbae or adults of G. boliviana were found in any of the fields and M. janthiniformis) have so far not reached or have not surveyed. In addition, a large open-field test was established been reported in areas where the respective nontarget plants in Argentina in 2000, exposing two popular varieties of occur (California). In the case of M. janthiniformis, eggplant to the beetles, with no feeding, oviposition, or nontarget attack on one native species, S. virga, is possible. development observed on eggplant (Gandolfo et al. 2007). For the three agents for which either prerelease open- The prediction therefore was that it would be highly field tests (Galerucella spp. and G. boliviana) or postrelease unlikely that G. boliviana would attack eggplant after open-field tests and field host range data (C. quadrigem- release (Gandolfo et al. 2007). mina) showed that nontarget attack did not have any major After 7 yr of postrelease monitoring, S. viarum has been economical or environmental effects, we believe that agents replaced by other plant species and no nontarget effects would be safe for release if petitioned today. For the two have been observed, even on congeners of the weed such as cases in which testing was inadequate, i.e., did not include the nonnative Solanum tampicense Dunal (wetland night- a sufficient number of native plants (M. janthiniformis)or shade) and S. torvum (Turkey berry) that were growing conclusive open-field test data (A. malherbae), results of intermixed with or in close proximity to S. viarum (Medal respective tests would need to be awaited to better define et al. 2008). Several thousand acres of eggplants, including the realized host range prior to making a decision. In sum, organic production, are being grown in Florida statewide, after reviewing all available data, it is our assessment that and there have been no reports of any damage or feeding the release of the discussed agents has resulted in a net on eggplant by G. boliviana, despite the statewide benefit to the U.S. economy and environment. distribution of the beetle in Florida (Diaz et al. 2014; J. Would They Be Released Today? We believe that for the Medal, personal communication). illustrated cases the agents would not be approved if Given that oviposition and development, though petitioned today. In three cases, this would be because of limited, occurred on a commercially important crop potential or confirmed attack on economically important (eggplant) in pre-release host-specificity tests, we conclude plant species (Hypericum calycinum, Lagerstroemia indica, that in the absence of criteria that specify tolerable risk Solanum viarum), and in two cases, because of potential levels, G. boliviana would not be permitted for release attack on rare, protected, or threatened and endangered today. plant species (Nuttallanthus canadensis and Calystegia spp.) (Table 1). We base our assessment on the reasoning Summary and Discussion provided for the denial of issuance of a release permit for C. basicorne (L. Smith, personal communication). Similar to Pre- vs. Postrelease Data for Our Five Case Studies. The the C. basicorne case, with regard to realized host-range biological control agents discussed in the five case studies data, two additional leaf beetles petitioned as biological were released in the United States as long as 68 and as control agents for tropical soda apple, Gratiana graminea recently as 12 yr ago. All have contributed to the sustainable Klug and Metriona elatior Klug, with a host range very and environmentally benign management of their respective similar to that of the released G. boliviana (Bredow et al. target weeds. For the three agents for which postrelease data 2007; Medal et al. 1999, 2010), were not recommended on nontarget attack was available, predictions of potential for release by the TAG (USDA APHIS 2014; J Medal, nontarget effects based on prerelease or postrelease host- personal communication). This is surprising, because G. specificity testing were accurate or overestimated attack graminea has a narrower fundamental host range than G. potential (C. quadrigemina for St. Johnswort, Galerucella boliviana with an average of 86 eggs laid on tropical soda spp. for purple loosestrife, and G. boliviana on tropical soda apple and only 0.1 on eggplant and no larval development apple). In addition, the documented nontarget attack was observed on eggplant whatsoever (Medal et al. 2010). transitory (decreased with time), decreased with distance from the weed infestation or following its successful control Improvement of Environmental Safety and Risk Stan- (e.g. Galerucella spp. for purple loosestrife; Corrigan et al. dards in Foreign Exploration and Biocontrol Regulation.

572 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 Table 1. Factors that could have prevented biological control agents discussed in case studies from being approved for release in the United States today, applying current policies and regulations. Year of release in the Attack on Attack on Objection by Target weed/biological control agent United States economic plant T&Ea species individual state St. Johnswort/Chrysolina quadrigemina 1945 X Field bindweed/Aceria malherbae 1989 X X Purple loosestrife/Galerucella spp. 1992 X X Dalmatian toadflax/Mecinus janthiniformis 1996 X X Tropical soda apple/Gratiana boliviana 2002 X

a Abbreviation: T&E, threatened and endangered.

We did not include the seed-head–feeding weevil Rhino- occurred in C. quadrigemina and M. janthiniformis (see cyllus conicus Fro¨lich, released in 1968 in the United States case studies), again illustrating the change in socio- for the control of invasive thistles in the genus Carduus,as economic values over time (Briese 2004). Even until the a case study in our paper, because it does not fit our early 1990s, the emphasis of host-specificity testing was to second selection criteria (i.e. that the agent would not be ensure that species of economic importance were not permitted for release today although it can be considered attacked by a potential agent even if these species were environmentally safe). Firstly, R. conicus has since its release completely unrelated to the target weed (also see Fowler et been found associated with the population decline of two al. 2004). Systematic inclusion of native North American native North American Cirsium species, Cirsium canescens species in test plant lists has dramatically changed in the Nutt. and Cirsium pitcheri (Torr. ex Eaton) Torr. & A. last 15 to 20 yr, partially in response to the Endangered Gray, listed as threatened in the United States (Louda et Species Act but also because of efforts to better predict the al. 2005a), and secondly, R. conicus was collected during realized host range in North America, using the actual original foreign exploration efforts from plant species in species that could be affected. Similarly, technological four different genera (Carduus, Cirsium, Onopordum and advancements that were unavailable at the time some Silybum;Zwo¨lfer 1967), clearly demonstrating that the agents were initially tested, especially the increasing weevil is oligophagous. availability of molecular genetic tools, have improved our The case of R. conicus illustrates how environmental understanding of phylogenetic relationships of target weeds safety standards in classical weed biological control have and native North American flora in many weed biological evolved since the 1960s. The National Environmental control programs (see Gaskin et al. 2011 for a review). Policy Act and the Endangered Species Act were signed Consequently, some released biological control agents were into law in 1970 and 1973, respectively, marking a societal never tested against the full range of native plant species shift in the perception of native fauna and flora and closely related to the target weed, and were introduced outlawing damage to threatened and endangered plant into the United States only to be found attacking native species in the United States. In hindsight, the release of R. species in later host-range tests (e.g., M. janthiniformis for conicus would be a clear violation of that law. However, the Dalmatian toadflax). In some instances in which native original host-specificity tests conducted for the weevil in plant species were attacked during tests, the respective the 1960s reflect the predominant agricultural values of the agents were still released, either because attack was judged time and consequently only included one native North to be transitory (e.g., Galerucella spp. for purple loosestrife) American Cirsium species, and that species was only or because benefits from control of the weed were judged to included in adult feeding trials (Gassmann and Louda outweigh the risk of harm to native plant species (e.g., 2001). The assessment of the potential effects of R. conicus Aceria malherbae for field bindweed; see Rosenthal and on native North American species was not a requirement Platts (1990). of testing procedures at the time (Gassmann and Louda 2001). Today, the seed-head weevil would not even be Consequences of Nontarget Attack. Even in cases where selected as a potential biological control agent, given its nontarget attack does occur after the release of a biological oligophagous feeding habit. Also, in 2014, it is standard control agent, this is not necessarily detrimental for the practice to only test a single source agent population rather nontarget plant on a population level. As two of our case than specimens collected from several host plant species studies (St. Johnswort and purple loosestrife) and numer- over a large spatial scale as was done with R. conicus. ous other studies have shown, nontarget attack is often Limited prerelease testing of native plant diversity also transitory and does not persist in time or ceases with

Hinz et al.: Successes we may not have had N 573

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 distance from agent outbreaks (Andres 1985; Blossey et al. Australia (Taylor et al. 2007; Q. Paynter, personal 2001; Julien and Griffiths 1998; McFadyen et al. 2002; communication). Schooler et al. 2003; Suckling and Sforza 2014). This Unfortunately, due to limited resources, postrelease kind of attack is often referred to as ‘‘spillover’’ and monitoring for nontarget attack is not systematically being is functionally a form of apparent competition, since it conducted and has mostly focused on the individual plant generally only occurs at high agent densities and in level (e.g., Taylor et al. 2007; Willis et al. 2003; but see presence of the target weed, i.e., the attack ceases with the Van Hezewijk et al. 2010; Weed and Schwarzla¨nder 2014). reduction of the target weed density. More long-term studies, such as the ones described above, A recent review on the magnitude of nontarget impacts are desperately needed to better understand if and how from classical weed biocontrol agents found that the vast direct nontarget attack results in negative or potentially majority of introduced agents (. 99%) have had no positive population-level consequences and to distinguish known significant adverse effect on nontarget plants thus sustained from transitory nontarget feeding. far (Suckling and Sforza 2014). For the cases for which impact could be quantified (n 5 140), and where impacts were detected, the majority (91.6%) were categorized as Recommendations minimal or minor in magnitude with no known adverse Develop Criteria for Risk and Benefit Assessments. In long-term impact on nontarget plant populations. The the United States, the Plant Protection Act does not specify only two cases in which moderate to massive negative direct guidelines on how to define and quantify risks of biological impacts occurred were from R. conicus on native North control agents to crops other than stating that a biological American Cirsium species (see above) and from the moth control agent ‘‘may not be a plant pest.’’ This limited Cactoblastic cactorum (Berg) (Suckling and Sforza 2014). legislative guidance has likely resulted in some of the Cactoblastic cactorum was introduced from South America confusion among researchers regarding the implementation to Australia in the 1920s and then to southern Africa, of policy for management activities such as biological where it contributed to the spectacular control of invasive control (also see Miller and Aplet 2005). The risks of Opuntia species (Zimmermann et al. 2010). From 1957 biological control are considered under multiple laws with onward, it was unfortunately also released on islands in the policies that emphasize the consideration and mitigation of Caribbean, from where it spread naturally or accidentally risk to crops (under the Plant Protection Act), risk to the through the plant-nursery trade to Florida. It is now environment (under the National Environmental Protec- threatening numerous native, economically used, and tion Act), or risk to threatened and endangered organisms culturally important Opuntia species in the southern (under the Endangered Species Act). Through the EA, the United States (Zimmermann et al. 2010). Certainly, the regulatory process as currently implemented also considers introduction of C. cactorum in the Caribbean would not be the risk of an invasive species itself to crops, the permitted today because of the risk of attack on nontarget environment, or threatened and endangered species as well native opuntias. as the consequences of ‘‘no action.’’ However, there exists However, there are also cases in which initial nontarget currently no formally outlined risk analysis including the attack resulted in a net positive effect for the nontarget likelihood of nontarget attack to occur and its magnitude species. The flea beetle Aphthona nigriscutis Foudras, in the evaluation process (Sheppard et al. 2003). In released for the control of leafy spurge, Euphorbia esula addition, USDA APHIS currently has no mandate to weigh L., attacked the native Euphorbia brachycera Engelm. these potential risks against the potential benefits of [5Euphorbia robusta (Engelm.) Small ex Britton & A. releasing the agent. Instead, if a potential risk to a crop, Br.], in mixed stands of both plants (Baker et al. 2004). In or any risk to a threatened and endangered species is found, the first year of postrelease monitoring (1999), 87% of E. the agent is most likely not being released. The selected case brachycera plants showed feeding damage by flea beetle studies illustrate that, historically, a broader interpretation adults and mortality of plants with heavy feeding was 36%. of regulations may have been possible. Initially, we However by 2002, when leafy spurge cover had declined recommend a reassessment of current policy to enable the from more than 50% to less than 6%, the E. brachycera consideration of both benefits and risks of all management population increased from 31 to 542 plants, of which less activities, including biological control and nonactions, than 3% showed any feeding damage. Hence, the focused at a habitat level, rather than for individual species biological control agent had a large net beneficial impact of concern as is done now. Ideally, such a reexamination on the native nontarget species and the indirect beneficial should include the formal addition of risk–benefit analyses effect increased with decreasing target weed densities. A with a ranking of decision-making criteria as part of the similar scenario has been described for the nontarget attack TAG’s evaluation of test plant lists when a biological of the moth Neurostrota gunniella Busck released for control project is initiated, rather than waiting until the control of Mimosa pigra L. on the native Neptunia major in process has moved to an EA consultation stage. This would

574 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 assist petitioners in the preparation of data sets for the the agent (see H. anagalloides and H. scouleri in St. petition; the agencies they represent, which would be able Johnswort case) (Andres 1985; Briese 2004). to allocate limited resources effectively; and the regulatory An approach that has been used to specifically validate agency, which could be assisted in the development of an prerelease host specificity data and predictions are common EA-like document by the petitioners and the agencies they gardens in the area of introduction, where test plant species represent. that were expected to be potentially attacked were exposed In the longer term we propose that legislation may be under seminatural conditions together with the respective amended to include risk–benefit analysis mechanisms in target weed to the agent (Center et al. 2007; Frye et al. the biocontrol introduction process in the United States as 2010; Pratt et al. 2009). Results of these experiments have has been done successfully in New Zealand and Australia shown that prerelease predictions on the risk of potential (Barratt and Moeed 2005; Cullen and Delfosse 1985; Hill nontarget attack were accurate to conservative, i.e., et al. 2013; Sheppard et al. 2003). Both exotic plant pests overestimating risk (Center et al. 2007; Frye et al. 2010; and biological weed control are regulated under the Pratt et al. 2009). These experiments are also much more authority of the Plant Protection Act and the jurisdiction cost-effective than large-scale postrelease monitoring sur- of USDA APHIS. A regulatory framework that addresses veys and can be conducted right after release, without simultaneously the management of exotic plant pests and waiting for the agent to establish and reach outbreak biological weed control could be developed that would densities. However, they do not replace postrelease provide great synergistic efficiencies and transparency, monitoring efforts (see below). and would enable science-based decisions (Sheppard et We believe that the incorporation of realized host range al. 2003). The procedural framework currently used in data from common garden or open-field experiments in the New Zealand under that nation’s Environmental Pro- country of origin needs to be recognized if not prioritized tection Authority most closely matches these criteria (for in the introduction process. These experiments should details see Barratt and Moeed 2005; Sheppard et al. explicitly be weighted more heavily in risk benefit analyses 2003). An important component of the New Zealand and release decisions for biological control agents. Environmental Protection Authority is that its decisions Multiple-choice experiments and studies explicitly investi- have a judicial basis, which cannot be overruled or gating the host selection behavior of candidate agents challenged (e.g., by lawsuits). Similar to New Zealand, consider the host-finding abilities and needs of candidate Australia passed a Biological Control Act in 1984, which biocontrol agents without which nontarget attack is legislates a regulatory process to ensure decisions in cases extremely unlikely to occur. If in the future only the where there are conflicting interests (Cullen and Delfosse fundamental host ranges of potential agents were to be 1985; McFadyen 1998). Common to both countries is considered as part of a zero-tolerance for risk decision- the creation of legislation specifically tailored toward making the number of available biological control agents classical biological control, which facilitates both the would decline dramatically (also see Sheppard et al. 2005). effective implementation and regulation of this manage- To our knowledge, there are no examples for biological ment tool. control agents that have a wider host range than that demonstrated in multiple-choice field tests (Blossey et al. Realized vs. Fundamental Host Range. It appears that for 2001; Clement and Cristofaro 1995; Paynter et al. 2004; the case of C. basicorne and the tropical soda apple agents Pratt et al. 2009). that were not recommended for release (G. graminea and M. elatior), results of open-field tests or their field host Formalize Postrelease Monitoring as Part of Permitting range in the native range (realized host range) were not Process. Multiple reviews of biological control have considered or weighted in importance relative to feeding, reiterated the need for postrelease monitoring to confirm oviposition, or development data under no-choice condi- effectiveness or lack thereof of released biocontrol agents, tions (fundamental host range). Both the case studies assess potential nontarget plant use and, generally to discussed in this paper and other experimental results validate prerelease host-range testing results (Blossey et al. (Center et al. 2007; Clement and Cristofaro 1995; Frye et 1994; Fowler et al. 2004; McFadyen 1998; Sheppard et al. al. 2010; Pratt et al. 2009) have emphasized the value of 2005). The only systematic nationwide survey for potential multiple-choice cage or open-field tests in the country of nontarget effects of weed biological control agents that we origin (prerelease) to accurately predict the realized host are aware of has been conducted in New Zealand, and it range of biological control agents postrelease. Even if covered 21 of the 30 agents established by 2004 in that nontarget plant species are attacked by an agent in country (Paynter et al. 2004; Waipara et al. 2009). experimental open-field tests, the plants are not necessarily Repeating a similar effort in the United States is unrealistic, utilized in the field, since lack of geographical or habitat considering the size of the country and the large number of overlap might prevent the co-occurrence of the plant and target weeds and released agents (118 agents released for 53

Hinz et al.: Successes we may not have had N 575

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 weeds at this time; R. Winston, personal communication). striking that the oversight of other management strategies However, smaller monitoring efforts, targeting an individ- for invasive plants does not routinely include consideration ual weed or group of related weeds or agent taxa, may be of nontarget or habitat-level effects, which have become feasible. Recent efforts to include all available nontarget common in biological control programs. The recom- data in the revised fifth edition of Biological Control of mended implementation of an overall risk–benefit ap- Weeds: A Worldwide Catalogue of Agents and Their proach would consider all management options using the Target Weeds (Winston et al. in press) is also a positive step same criteria and result in better decision-making. The forward. reviewed case studies and regulatory systems for biological From a regulatory point of view we recommend a weed control in other countries demonstrate that it is mandated requirement for organizations granted release possible to reliably balance the risks, costs, and benefits of permits to conduct follow-up studies on release impacts biological weed control. In a time of ever-shrinking and first-order nontarget interactions. Postrelease monitor- resources, biological control remains one of the more ing, including that of nontarget attack, is currently required cost-effective and the only truly sustained means for the in every release permit granted; however, the extent to management of exotic invasive plants in the United States. which that monitoring occurs is not always clear. Thus, we It would be unfortunate to lose biological control as an recommend development of a standardized reporting option in our management toolbox. system that requires the releasing organizations to provide follow-up monitoring data in the public domain. Better postrelease data would also help test the case for the safety Acknowledgments and efficacy of biological control as a management strategy, This article is a based on a paper presented by H. L. Hinz, address concerns about the lack of data on the fate of exotic A. Gassmann, and M. Schwarzla¨nder during the 2nd insect releases, and facilitate evaluations of future agents. Conference of the Northern Rockies Invasive Species Council (NRIPC), on October 27, 2010, in Coeur d’Alene, ID, and subsequent discussion among the authors. We would like Conclusions to thank Rose DeClerck-Floate (Agriculture and Agri-Food Canada, Lethbridge, Canada), Richard Hansen (USDA, Biological weed control has significantly advanced as a APHIS, Center for Plant Health Science Technology, Ft. scientific discipline since the 1950s when Chrysolina was Collins, CO), Quentin Paynter (Landcare Research, Auck- released for the control of St. Johnswort and subsequently land, New Zealand), Mike Pitcairn (California Department honored for its success by a monument raised in California. of Food and Agriculture, Sacramento, CA), Lincoln Smith But we recognize that biological control is an irreversible (USDA Agricultural Research Service, Albany, CA), and action and for that reason there is such careful review and especially Robert Tichenor (USDA APHIS Plant Protection oversight of the decision to release exotic organisms in and Quarantine, Riverdale, MD) and two anonymous all countries in the world that utilize biological control. reviewers for helpful information and critically reviewing A consistent argument in favor of other management earlier versions of the manuscript. activities such as mechanical control or herbicides is that they are reversible whereas biological control is not. Although it is true that other management activities can Literature Cited be halted, it is certainly not true that if they are stopped, Albach DC, Meudt HM, Oxelman B (2005) Piecing together the ‘‘new’’ the habitat will revert to some sort of better, let alone Plantaginaceae. Am J Bot 92:297–315 native, state. Once an invasive plant species has moved into Andres LA (1985) Interactions of Chrysolina quadrigemina and Hypericum species in California. Pages 235–239 in Proceedings of the habitat, restoration is extremely difficult and the habitat the VI International Symposium on Biological Control of Weeds. is often irrevocably changed. Herbicides themselves can Ottawa, Canada: Agriculture Canada have longer-term residual effects that, for instance, limit the Baker JL, Webber NAP, Johnson KK (2004) Non-target impacts of regeneration of native plants (e.g., Wagner and Nelson Aphthona nigriscutis, a biological control agent for Euphorbia esula 2014). Biological weed control, conducted with prudence, (leafy spurge), on a native plant Euphorbia robusta. Pages 247–251 in Proceedings of the XI International Symposium on Biological overall has a long history of safe and effective implemen- Control of Weeds. Canberra, Australia: CSIRO Entomology tation around the world (Blossey et al. 2001; McFadyen Barratt BIP, Moeed A (2005) Environmental safety of biological 1998; Paynter et al. 2004; Pemberton 2000; Suckling and control: policy and practice in New Zealand. Biol Control 35: Sforza 2014). Similarly, herbicides can be effective if used 247–252 appropriately. A key difference between management Blossey B, Casagrande R, Tewksbury L, Landis DA, Wiedenmann RN, strategies appears to be that biological control is being Ellis DR (2001) Nontarget feeding of leaf-beetles introduced to control purple loosestrife (Lythrum salicaria L.). Nat Areas J 21: held to a higher ecological impact standard than other 368–377 control tactics, with recent emphases on nontarget effects Blossey B, Schroeder D, Hight SD, Malecki RA (1994) Host specificity or secondary trophic effects or even food web effects. It is and environmental impact of two leaf beetles (Galerucella calmariensis

576 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 and G. pusilla) for biological control of purple loosestrife (Lythrum Pages 265–270 in Proceedings of the XI International Symposium on salicaria). Weed Sci 42:134–140 Biological Control of Weeds. Canberra, Australia: CSIRO Entomology Boldt PE, Rosenthal SS, Srinivasan R (1998) Distribution of field Frye MJ, Lake EC, Hough-Goldstein J (2010) Field host-specificity of bindweed and hedge bindweed in the USA. J Prod Agric 11:377–381 the mile-a-minute weevil, Rhinoncomimus latipes Korotyaev (Coleop- Bredow E, Pedrosa-Macedo JH, Medal JC, Cuda JP (2007) Open field tera: Curculionidae). Biol Control 55:234–240 host specificity tests in Brazil for risk assessment of Metriona elatior Gandolfo D, McKay F, Medal JC, Cuda JP (2007) Open-field host (Coleoptera: Chrysomelidae), a potential biological control agent for specificity test of Gratiana boliviana (Coleoptera: Chrysomelidae), a Solanum viarum (Solanaceae) in Florida. Fla Entomol 90:559–564 biological control agent of tropical soda apple (Solanaceae) in the Breiter N, Seastedt TR (2007) Postrelease evaluation of Mecinus United States. Fla Entomol 90:223–228 janthinus host specificity, a biological control agent for invasive Gaskin JF, Bon MC, Cock MJW, Cristofaro M, De Biase A, De Clerck- toadflax (Linaria spp.). Weed Sci. 55:164–168 Floate R, Ellison CA, Hinz HL, Hufbauer RA, Julien MH, Sforza R Briese DT (2004) Weed biological control: applying science to solve (2011) Applying molecular-based approaches to classical biological seemingly intractable problems. Aust J Ecol 43:304–317 control of weeds. Biol Control 58:1–21 Center TD, Pratt PD, Tipping PW, Rayamajhi MB, Van TK, Wineriter Gassmann A, Louda SM (2001) Rhinocyllus conicus: initial evaluation SA, Dray, Jr., FA (2007) Initial impacts and field validation of host and subsequent ecological impacts in North America. Pages 147–183 range for Boreioglycaspis melaleucae Moore (Hemiptera: Psyllidae), a in Wajnberg E, Scott JK, Quimby PC, eds. Evaluating Indirect biological control agent of the invasive tree Melaleuca quinquenervia Ecological Effects of Biological Control. Wallingford, UK: CABI (Cav.) Blake (Myrtales: Myrtaceae: Leptospermoidaea). Environ Publishing Entomol 36:569–576 Gassmann A, Tosˇevski I, Speiwak E, Heinlein G, Rechner H, Yaworski Clement SL, Cristofaro M (1995) Open-field tests in host-specificity K, Thalmann M (2002) Biological Control of Toadflaxes, Linaria determination of insects for biological control of weeds. Biocontrol spp. Annual Report 2001. Dele´mont, Switzerland: CABI Bioscience Sci Technol 5:395–406 Switzerland Centre. 15 p Cofrancesco AF, Shearer JF (2004) Technical Advisory Group for Groenteman R, Fowler SV, Sullivan JJ (2011) St. John’s wort beetles Biological Control Agents of Weeds. Pages 38–41 in Coombs EM, would not have been introduced to New Zealand now: a retrospective Clark JK, Piper GL, Cofrancesco AF, eds. Biological Control of host range test of New Zealand’s most successful weed biocontrol Invasive Plants in the United States. Corvallis, OR: Oregon State agents. Biol Control 57:50–58 University Press Hill R, Campbell D, Hayes L, Corin S, Fowler S (2013) Why the New Corrigan JE, Mackenzie DL, Simser L (1998) Field observations of non- Zealand regulatory system for introducing new biological control target feeding by Galerucella calmariensis (Coleoptera: Chrysome- agents works. Pages 75–83 in Proceedings of the XIII International lidae), an introduced biological control agent of purple loosestrife, Symposium on Biological Control of Weeds. Morgantown, West Lythrum salicaria (Lythraceae). Pages 99–106 in Proceedings of the Virginia: USDA Forest Service Forest Health Technology Enterprise Entomological Society of Ontario Vol. 129. Guelph, Ontario: Team Entomological Society of Ontario Holloway JK (1948) Biological control of Klamath weed. Progress Cristofaro M, De Biase A, Smith L (2013) Field release of a prospective report. Econ Entomol 41:56–57 biological control agent of weeds, Ceratapion basicorne, to evaluate Huffaker CB (1967) A comparison of the status of biological weed potential risk to a nontarget crop. Biol Control 64:305–314 control of St. Johnswort in California and Australia. Mushi Cullen JM, Delfosse ES (1985) Echium plantigineum: catalyst for 39(suppl.):51–73 conflict and change in Australia. Pages 249–292 in Proceedings of the Huffaker CB, Kennett CE (1959) A ten-year study of vegetational VI International Symposium on Biological Control of Weeds. changes associated with biological control of Klamath weed. J Range Ottawa, Canada: Agriculture Canada Manag 12:69–82 Currie GA, Garthside S (1932) The possibility of the entomological Jeanneret P, Schroeder D (1992) Biology and host specificity of Mecinus control of St. John’s Wort in Australia. Melbourne, Australia: janthinus Germar (Col.: Curculionidae), a candidate for the biological Progress Rep. Council for Scientific and Industrial Research 29. 28 p control of yellow and Dalmatian toadflax, Linaria vulgaris (L.) Mill. DeBach P, Rosen D (1991) Biological control by natural enemies. and Linaria dalmatica (L.) Mill. (Scrophulariaceae) in North Cambridge, UK: Cambridge University Press. 211 p America. Biocontrol Sci Technol 2:25–34 De Clerck-Floate RA, Harris P (2002) Linaria dalmatica (L.) Miller, Jensen KIN, Harris P, Sampson MG (2002) Hypericum perforatum L., Dalmatian toadflax (Scrophulariaceae) Pages 368–374 in Mason PG, St John’s wort (Clusiaceae). Pages 361–368 in Mason PG, Huber JT, Huber JT, eds. Biological Control Programmes in Canada, 1981– eds. Biological Control Programmes in Canada, 1981–2000. Wall- 2000. Wallingford, UK: CABI Publishing ingford, UK: CABI Publishing Diaz R, Manrique V, Hibbard K, Fox AM, Roda A, Gandolfo D, Julien MH, Griffiths MW (1998) Biological Control of Weeds. A World McKay F, Medal J, Hight S, Overholt WA (2014) Successful Catalogue of Agents and their Target Weeds. 4th edn. Wallingford, biological control of tropical soda apple (Solanales: Solanaceae) in UK: CAB International. 223 p Florida: a review of key program components. Fla Entomol 1: Kok LT, McAvoy TJ, Malecki RA, Hight SD, Drea JJ, Coulson JR 179–190 (1992) Host specificity tests of Galerucella calmariensis (L.) and G. Dudley TL, Kazmer DJ (2005) Field assessment of the risk posed by pusilla (Duft.) (Coleoptera: Chrysomelidae), potential biological Diohabda elongata, a biocontrol agent for control of saltcedar control agents of purple loosestrife, Lythrum salicaria L. (Lythraceae). (Tamarix spp.), to a nontarget plant Frankenia salina. Biol Control Biol Control 2:282–290 35:265–275 Landis DA, Sebolt DC, Haas MJ, Klepinger M (2003) Establishment Ferna´ndez-Mazuecos M, Blanco-Pastor J, Vargas P (2013) A phylogeny and impact of Galerucella calmariensis L. (Coleoptera: Chrysomelidae) of toadflaxes (Linaria Mill.) based on nuclear internal transcribed on Lythrum salicaria L. and associated plant communities in spacer sequences: systematic and evolutionary consequences. Int J Plant Michigan. Biol Control 28:78–91 Sci 174:234–249 Lindgren CJ, Corrigan J, De Clerck-Floate RA (2002) Lythrum salicaria Fowler SV, Gourlay AH, Hill RH, Withers T (2004) Safety in New L., purple loosestrife (Lythraceae). Pages 383–390 in Mason PG, Zealand weed biocontrol: a retrospective analysis of host-specificity Huber JT, eds. Biological Control Programmes in Canada, 1981– testing and the predictability of impacts on non-target plants. 2000. Wallingford, UK: CABI Publishing

Hinz et al.: Successes we may not have had N 577

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 Littlefield JL (2004) Bindweeds. Pages 150–151 in Coombs EM, Clark Wiggins L, Kirby C, Crawford S (2009) Releases, distribution and JK, Piper GL, Cofrancesco AF Jr, eds. Biological Control of Invasive abundance of Gratiana boliviana (Coleoptera: Chrysomelidae), a Plants in the United States. Corvallis, Oregon: Oregon State biological control agent for tropical soda apple (Solanum viarum, University Press Solanaceae), in Florida. Fla Entomol 92:450–457 Louda SM, Rand TA, Arnett AE, McClay AS, Shea K, McEachern AK Overholt WA, Diaz R, Markle L, Medal JC (2010) The effect of (2005a) Evaluation of ecological risk to populations of a threatened Gratiana boliviana (Coleoptera: Chrysomelidae) herbivory on growth plant from an invasive biocontrol insect. Ecol Appl 15:234–249 and population density of tropical soda apple (Solanum viarum)in Louda SM, Rand TA, Russell FL, Arnett AE (2005b) Assessment of Florida. Biocontrol Sci Technol 20:791–807 ecological risks in weed biocontrol: input from retrospective Paynter Q, Fowler SV, Gourlay AH, Haines ML, Harman HM, Hona ecological analyses. Biol. Control 35:253–264 SR, Peterson PG, Smith LA, Wilson-Davey JRA, Winks CJ, Withers McClay AS, De Clerck-Floate RA (2002a) Convolvulus arvensis L., field TM (2004) Safety in New Zealand weed biocontrol: a nationwide bindweed (Convolvulaceae). Pages 331–337 in Mason PG, Huber JT, survey for impacts on non-target plants. NZ Plant Protect 57: eds. Biological Control Programmes in Canada, 1981–2000. Wall- 102–107 ingford, UK: CABI Publishing Pemberton RW (2000) Predictable risk to native plants in weed McClay AS, De Clerck-Floate RA (2002b) Linaria vulgaris Miller, biological control. Oecologia 125:489–494 Yellow Toadflax (Scrophulariaceae). Pages 375–382 in Mason PG, Piper GL (2004) St. Johnswort. Pages 322–334 in Coombs EM, Clark Huber JT, eds. Biological Control Programmes in Canada, 1981– JK, Piper GL, Cofrancesco AF Jr, eds. Biological Control of Invasive 2000. Wallingford, UK: CABI Publishing Plants in the United States. Corvallis, Oregon: Oregon State McClay AS, De Clerck-Floate RA (2013) Convolvulus arvensis L., field University Press bindweed (Convolvulaceae). Pages 307–309 in Mason PG, Gillespie Piper GL, Coombs EM, Blossey B, McEvoy PB, Schooler SS (2004) DR, eds. Biological Control Programmes in Canada, 2001–2012. Purple loosestrife. Pages 281–292 in Coombs EM, Clark JK, Piper Wallingford, UK: CABI Publishing GL, Cofrancesco AF Jr, eds. Biological Control of Invasive Plants in McFadyen REC (1998) Biological control of weeds. Ann Review the United States. Corvallis, Oregon: Oregon State University Press Entomol 43:369–393 Pratt PD, Rayamajhi MB, Center TD, Tipping PW, Wheeler GS McFadyen RE, Vitelli M, Setter C (2002) Host specificity of the rubber (2009) The ecological host range of an intentionally introduced vine moth, Euclasta whalleyi Popescu-Gorj and Constantinescu herbivore: a comparison of predicted versus actual host use. Biol. (Lepidoptera: Crambidae: Pyraustinae): field host-range compared Control 49:146–153 to that predicted by laboratory tests. Aust J Entomol 41:321–323 Rector BG, De Biase A, Cristofaro M, Primerano S, Belvedere S, Medal J, Bustamante N, Vitorino M, Beal L, Overholt W, Diaz R, Cuda Antonini G, Sobhian R (2010) DNA Fingerprinting to improve data J (2010) Host specificity tests of Gratiana graminea (Coleoptera: collection efficiency and yield in an open-field host-specificity test of a Chrysomelidae), a potential biological control agent of tropical soda weed biological control candidate. Invasive Plant Sci Manag 3: apple, Solanum viarum (Solanaceae). Fla Entomol 93:231–242 429–439 Medal JC, Cuda JP (2010) Establishment and initial impact of the leaf- Rosenthal SS, Platts BE (1990) Host specificity of Aceria (Eriophyes) beetle Gratiana boliviana (Chrysomelidae), first biocontrol agent malherbe, [Acari: Eriophyidae], a biological control agent for the weed, released against tropical soda apple in Florida. Fla Entomol 93: Convolvulus arvensis [Convolvulaceae]. Entomophaga 35:459–463 493–500 Salaudeen T, Thomas M, Harding D (2012) II. Economics of tropical Medal J, Overholt WA, Stansly P, Roda A, Osborne L, Hibbard K, soda apple in Florida. Pages 10–19 in Medal J, Overholt W, Charudattan Gaskalla R, Burns E, Chong J, Sellers B, Hight SD, Cuda JP, R, Mullahey J, Gaskalla R, Dı´az R, Cuda J, eds. Tropical Soda Apple Vitorino M, Bredow E, Pedrosa-Macedo JH, Wikler C (2008) Management Plan. Gainesville, FL: University of Florida-IFAS. Florida Establishment, spread and initial impacts of Gratiana boliviana Department of Agriculture and Consumer Services-DPI. 87 p (Chrysomelidae) on Solanum viarum in Florida. Pages 389–393 in Schaffner U (2001) Host range testing of insects for biological weed Proceedings of the XII International Symposium on Biological control: how can it be better interpreted? BioScience 51:951–959 Control of Weeds. Wallingford, UK: CABI International Schat M, Sing SE, Peterson RKD, Menalled FD, Weaver DK (2011) Medal JC, Pitelli RA, Santana A, Gandolfo D, Gravena R, Habeck DH Growth inhibition of Dalmatian toadflax, Linaria dalmatica (L.) (1999) Host specificity of Metriona elatior, a potential biological Miller, in response to herbivory by the biological control agent control agent of tropical soda apple, Solanum viarum, in the USA. Mecinus janthinus Germar. J Entomol Sci 46:232–246 BioControl 44:421–436 Schooler SS, Coombs EM, McEvoy P (2003) Nontarget effects on crepe Medal JC, Sudbrink D, Gandolfo D, Ohashi D, Cuda JP (2002) myrtle by Galerucella pusilla and G. calmariensis (Chrysomelidae), Gratiana boliviana, a potential biocontrol agent of Solanum viarum: used for biological control of purple loosestrife (Lythrum salicaria). quarantine host-specificity testing in Florida and field surveys in Weed Sci 51:449–455 South America. BioControl 47:445–461 Sheppard AW, Hill R, DeClerck-Floate RA, McClay A, Olckers T, Miller ML, Aplet GH (2005) Applying legal sunshine to the hidden Quimby PC Jr, Zimmermann HG (2003) A global review of risk- regulation of biological control. Biol Control 35:358–365 benefit-cost analysis for the introduction of classical biological control Mullahey JJ, Mislevy P, Brown WF, Kline WN (1996) Tropical soda agents against weeds: a crisis in the making? Biocontrol News Inf 24: apple, an exotic weed threatening agriculture and natural systems. 91N–108N Dow Elanco. Down to Earth Vol. 51, No. 1. 8 p Sheppard AW, van Klinken RD, Heard TA (2005) Scientific advances Mullahey JJ, Nee M, Wunderlin RP, Delaney KR (1993) Tropical soda in the analysis of direct risks of weed biological control agents to apple (Solanum viarum) : a new weed threat in subtropical regions. nontarget plants. Biol Control 35:215–226 Weed Technol 7:783–786 Smith JM (1958) Biological control of Klamath weed, Hypericum Nowierski RM (2004) Toadflax. Pages 379–395 in Coombs EM, Clark perforatum L., in British Columbia. Pages 561–565 in Proceedings of JK, Piper GL, Cofrancesco AF Jr, eds. Biological Control of Invasive the 10th International Congress of Entomology. Ottawa, Canada: Plants in the United States. Corvallis, Oregon: Oregon State Mortimer University Press Smith L (2007) Physiological host range of Ceratapion basicorne,a Overholt WA, Diaz R, Hibbard K, Roda A, Amalin D, Fox A, Hight prospective biological control agent of Centaurea solsticialis (Aster- SD, Medal J, Stansly P, Carlisle B, Walter J, Hogue P, Gary L, aceae). Biological Control 41:120–133

578 N Invasive Plant Science and Management 7, October–December 2014

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1 Smith L, de Lillo E, Amrine JW Jr (2010) Effectiveness of eriophyid Villegas B, Gibbs C, Donnelly J (2011) A case of fortuitous biological mites for biological control of weedy plants and challenges for future control of Dalmatian toadflax in Susanville. Pages 64–66 in Woods research. Exp Appl Acarol 51:115–149 DM, ed. Biological Control Program 2010 Annual Summary. Smith L, Hayat R, Cristofaro M, Tronci C, Tozlu G, Lecce F (2006) Sacramento, California: California Department of Food and Assessment of risk of attack to safflower by Ceratapion basicorne Agriculture Plant Health and Pest Prevention Services (Coleoptera: Apionidae), a prospective biological control agent of Wagner V, Nelson CR (2014) Herbicides can negatively affect seed Centaurea solstitialis (Asteraceae). Biol Control 36:337–344 performance in native plants. Restor Ecol 22:288–291 Suckling DM, Sforza RFH (2014) What magnitude are observed non- Waipara N, Barton J, Smith LA, Harman HM, Winks C, Massey B, target impacts from weed biocontrol? PLOS ONE 9:1–12 Wilkie JP, Gianotti A, Cripps MG (2009) Safety in New Zealand Taylor DBJ, Heard TA, Paynter Q, Spafford H (2007) Nontarget effects weed biocontrol: a nationwide pathogen survey for impacts on non- of a weed biological control agent on a native plant in Northern target plants. NZ Plant Protect 62:41–49 Australia. Biol Control 42:25–33 Wapshere AJ (1989) A testing sequence for reducing rejection of Tosˇevski I, Caldara R, Jovic´ J, Hernandez-Vera G, Baviera C, Gassmann potential biological control agents for weeds. Ann Appl Biol 114: A, Emerson BC (2011) Morphological, molecular and biological 515–526 evidence reveal two cryptic species in Mecinus janthinus Germar Ward SM, Fleischmann CE, Turner MF, Sing SE (2009) Hybridization (Coleoptera, Curculionidae), a successful biological control agent of between invasive populations of Dalmatian toadflax (Linaria Dalmatian toadflax, Linaria dalmatica (Lamiales, Plantaginaceae). dalmatica) and yellow toadflax (Linaria vulgaris). Invasive Plant Sci Syst Entomol 36:741–753 Manag 2:369–378 Tosˇevski I, Mitrovic´ M, Krstic´ O, Jovic´ J, Gassmann A (2012) Weed AS, Schwarzla¨nder M (2014) Density dependence, precipitation Biological Control of Dalmatian and Yellow Toadflaxes (Linaria and biological control agent herbivory influence landscape-scale ´ dalmatica and L. vulgaris). Annual Report 2011. Delemont, dynamics of the invasive Eurasian plant Linaria dalmatica. J Appl Switzerland: CABI Bioscience Switzerland Centre. 69 p Ecol 51:825–834 USDA. 2000. Reviewer’s Manual for the Technical Advisory Group for Willis AJ, Berentson PR, Ash JE (2003) Impacts of a weed biocontrol Biological Control Agents of Weeds. Guidelines for Evaluating the agent on recovery from water stress in a target and non-target Safety of Candidate Biological Control Agents. United States Hypericum species. J Appl Ecol 40:320–333 Department of Agriculture, Marketing and Regulatory Programs, Winston RL, Schwarzla¨nder M, Hinz HL, Day MD, Cock MJW, Julien Animal and Plant Health Inspection Service, Plant Protection and Quarantine, 1st Edition. MH (in press) Biological Control of Weeds: A World Catalogue of USDA APHIS (2014) Technical Advisory Group for Biological Control Agents and their Target Weeds. 5th edn. MorgantownWest Virginia: Agents of Weeds TAG Petitions – APHIS Actions. http://www.aphis. USDA Forest Service, Forest Health Technology Enterprise Team. usda.gov/plant_health/permits/tag/downloads/TAGPetitionAction.pdf. FHTET-2014-XX. 223 p Accessed: November 13, 2014 Wilson LM, Sing SE, Piper GL, Hansen RW, De Clerck-Floate RA, Van Hezewijk BH, Bourchier RS, De Clerck-Floate RA (2010) MacKinnon DK, Randall C (2005) Biology and biological control of Regional-scale impact of the weed biocontrol agent Mecinus janthinus Dalmatian and yellow toadflax. Morgantown, West Virginia: USDA on Dalmatian toadflax (Linaria dalmatica). Biol Control 55:197–202 Forest Service, Forest Health Technology Enterprise Team-05-13. Van Klinken RD, Edwards OR (2002) Is host-specificity of weed 116 p biological control agents likely to evolve rapidly following establish- Zimmermann HG, Moran VC, Hoffmann JH (2000) The renowned ment? Ecol Lett 5:590–596 cactus moth, Cactoblastis cactorum: its natural history and threat to Van Wilgen BW, Moran VB, Hoffmann JH (2013) Some perspectives native Opuntia floras in Mexico and the United States of America. on the risks and benefits of biological control of invasive alien plants Divers Distrib 6:259–269 in the management of natural ecosystems. Environ Manag 52: Zwo¨lfer H (1967) The Host Range, Distribution, and Life-History of 531–540 Rhinocyllus conicus Fro¨l. (Col, Curculionidae). Progress Report. Villegas B (2009) Biological control of Dalmatian toadflax in California. Dele´mont, Switerland: CAB International Institute of Biological Pages 59–60 in Woods DM, ed. Biological Control Program 2008 Control. 18 p Annual Summary. Sacramento, California: California Department of Food and Agriculture Plant Health and Pest Prevention Services Received November 21, 2013, and approved July 16, 2014.

Hinz et al.: Successes we may not have had N 579

Downloaded from https://www.cambridge.org/core. IP address: 170.106.40.139, on 27 Sep 2021 at 05:35:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1614/IPSM-D-13-00095.1